Lysosomes degrade macromolecules and recycle their nutrient content to support cell function and survival. Yet, the machineries involved in lysosomal recycling of many nutrients remain to be discovered, with a notable example being choline, an essential metabolite liberated via phospholipid degradation. Here, we engineered metabolic dependency on lysosome-derived choline in pancreatic cancer cells to perform an endolysosome-focused CRISPR-Cas9 screen for genes mediating lysosomal choline recycling. We identified the orphan lysosomal transmembrane protein SPNS1 as critical for cell survival under choline limitation. SPNS1 loss leads to intralysosomal accumulation of lysophosphatidylcholine (LPC) and lysophosphatidylethanolamine (LPE). Mechanistically, we reveal that SPNS1 is a proton gradient-dependent transporter of LPC species from the lysosome for their re-esterification into phosphatidylcholine in the cytosol. Finally, we establish that LPC efflux by SPNS1 is required for cell survival under choline limitation. Collectively, our work defines a lysosomal phospholipid salvage pathway that is essential under nutrient limitation, and more broadly, provides a robust platform to deorphan lysosomal gene function.

Endolysosomal library construction

A list of lysosomal, endocytic and autophagy-related genes in addition to a selective set of metabolism and signaling genes (1,061 genes) was manually curated by unbiased profiling of lysosomal proteins using in-house LysoIP proteomics and by selecting genes implicated in endolysosomal trafficking. For the library design, 10 targeting sgRNAs were selected per gene, and an additional 1,050 sgRNAs (~9% of the library) were included as safe-targeting (control) sgRNAs. These guides were cloned into pMCB320.

Endolysosomal screen

The endolysosomal library was infected into Cas9-expressing PaTu-8988T pancreatic cancer cells at 1000x library coverage and selected using puromycin. At time point 0, cells were trypsinized, washed, and split into two culture media conditions: choline-depleted medium (-choline) or choline-supplemented medium (+choline). Choline-depleted medium was comprised of 1:1 DMEM/Ham’s F12 without sodium bicarbonate, serine, methionine, or choline chloride (Caisson Labs; DFP14-5) supplemented with 1200 mg/L sodium bicarbonate, 18 mg/L methionine, 27 mg/L serine, 5% triple-dialyzed FBS and penicillin/streptomycin. To make +choline medium, -choline medium was supplemented with 100 μM choline chloride. 

Cells were propagated in -choline or +choline medium for 14 doublings while maintaining a library coverage of >1000x. DNA from time point 0 (T0) and after fourteen doublings for each condition was isolated and sgRNA sequences were PCR-amplified. Illumina universal adaptor sequences were subsequently added to sgRNA amplicons by PCR, and sgRNA abundances were determined using the Illumina NextSeq 550 with NextSeq 500/550 mid-output kit v2.5.

Screen analysis

Sequencing data were analyzed using the CasTLE method developed by Morgens et al. In brief, the method calculates the most likely maximum effect (phenotype) size (CasTLE effect) among each group of gene-targeting sgRNAs by comparing each set to the negative sgRNAs (control and safe-targeting) in the library. The method then scores the significance of the effect (CasTLE score) by permuting the results.